Skip to main content

Pose Estimation on Jetson

This section demonstrates human pose estimation on NVIDIA Jetson using Ultralytics YOLO Pose models. Examples include image, video, and real-time camera inference.

1. Enable Optimal Jetson Performance​

Before running pose estimation, ensure Jetson is running at maximum performance.

Enable MAX Power Mode​

sudo nvpmodel -m 2

Enable Jetson Clocks​

sudo jetson_clocks

2. Pose Estimation on Images​

Enter Demo Directory​

cd ~/ultralytics/ultralytics/yahboom_demo

Run Image Pose Estimation​

python3 03.pose_image.py

Results are saved to:

~/ultralytics/ultralytics/output/

Sample Code (Image Pose Estimation)​

from ultralytics import YOLO

model = YOLO("yolo11n-pose.pt")
results = model("assets/people.jpg")

for r in results:
    r.show()
    r.save(filename="output/people_pose_output.jpg")

3. Pose Estimation on Video​

Run Video Pose Estimation​

python3 03.pose_video.py

Output video location:

~/ultralytics/ultralytics/output/

Sample Code (Video Pose Estimation)​

import cv2
from ultralytics import YOLO

model = YOLO("yolo11n-pose.pt")
cap = cv2.VideoCapture("videos/people.mp4")

width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = int(cap.get(cv2.CAP_PROP_FPS))

out = cv2.VideoWriter(
    "output/people_pose_output.mp4",
    cv2.VideoWriter_fourcc(*"mp4v"),
    fps,
    (width, height)
)

while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break

    results = model(frame)
    annotated = results[0].plot()
    out.write(annotated)

cap.release()
out.release()

4. Real-Time Pose Estimation​

  • USB Camera: OpenCV VideoCapture(0)
  • CSI Camera: GStreamer pipeline (nvarguscamerasrc)

Real-time processing follows the same inference logic as video pose estimation.

5. Notes​

  • Pose models output keypoints for each detected person
  • Suitable for motion tracking and activity analysis
  • Use Nano pose models for real-time inference
  • Export models to TensorRT for production deployment

Maintained by HemiHex for Jetson-based advanced vision workflows.